Merge branch 'feature/nonlinear-sumfact-cg' into 'master'

ref:dominic/dune-perftool Contains [!264] and master\....

See merge request [dominic/dune-perftool!269]

  [!264]: gitlab.dune-project.org/NoneNone/merge_requests/264
  [dominic/dune-perftool!269]: gitlab.dune-project.org/dominic/dune-perftool/merge_requests/269

Merge branch 'feature/sum-factorization-gmsh-files' into 'master'

ref:dominic/dune-perftool On unstructured grids we need to make sure that the
quadrature points on intersections on neighboring cells are visited in the
same order. This is done using edge consistent grids and by constructing the
geometry mappings in the right way.

Still missing:

-   Vectorization
-   Make UG grids edge consistent without writing and reading a gmsh file

See merge request [dominic/dune-perftool!264]

  [dominic/dune-perftool!264]: gitlab.dune-project.org/dominic/dune-perftool/merge_requests/264

Needed in dune-perftool-playground for Navier stokes

There seems to still be some problem.

Merge branch 'feature/better-subdomain-handling' into 'master'

See merge request [dominic/dune-perftool!268]

  [dominic/dune-perftool!268]: Nonedominic/dune-perftool/merge_requests/268

Note: Not yet working for gmsh files

Merge branch 'feature/lambda-generation-with-more-geometry' into 'master'

Target: normal-dependent constraints assembly.

See merge request [dominic/dune-perftool!267]

  [dominic/dune-perftool!267]: gitlab.dune-project.org/dominic/dune-perftool/merge_requests/267

Merge branch 'feature/fix-constraints-assembly-bug' into 'master'

We previously used bool() on the constraints expression to find out whether we
need to use the DIrichletConstraintsAssembler. However, UFL does the stupid
thing of evaulating

`bool(eq(x[0], 0))`

to False, because x[0] and 0 are obvioulsy not the same. Veeeery bad behaviour
for a symbolic language.

See merge request [dominic/dune-perftool!266]

  [dominic/dune-perftool!266]: gitlab.dune-project.org/dominic/dune-perftool/merge_requests/266

We previously used bool() on the constraints expression to find
out whether we need to use the DIrichletConstraintsAssembler.
However, UFL does the stupid thing of evaulating

bool(eq(x[0], 0))

to False, because x[0] and 0 are obvioulsy not the same. Veeeery
bad behaviour for a symbolic language.

Merge branch 'feature/small-fix-for-conditional' into 'master'

See merge request [dominic/dune-perftool!265]

  [dominic/dune-perftool!265]: Nonedominic/dune-perftool/merge_requests/265

This is done by throwing an exception in the 3D case and doing the 2D case in
the catch part.

Merge branch 'feature/costmodel-validation' into 'master'

The MR adds a new vectorization strategy target, which tries to find a
vectorization strategy whose cost (according to the model strategy) is as
close as possible to a given target. This allows to run several executables
and plot the cost model value against performance data. The actual plotting
work is done in dune-perftool-paperplots. The feature is very niche, and I
tried to minimize the occurence of code paths only necessary for this, but I
think the outcome is quite okay.

See merge request [dominic/dune-perftool!262]

  [dominic/dune-perftool!262]: gitlab.dune-project.org/dominic/dune-perftool/merge_requests/262

We do this from multiple subdirs